Botnets Abhishek Debchoudhury Jason Holmes What is a botnet? A network of computers running software that runs autonomously. In a security context we are interested in botnets in which the computers have been compromised and are under the control of a malicious adversary. What are botnets used for? • Spam o ~85% of email is spam • DDoS attacks • Identity theft o Cost in 2006: $15.6 billion • Phishing attacks o 4500 active sites at any given time, 1 million previously active sites What are botnets used for? • Hosting pirated software • Hosting and distributing malware • Click fraud o ~14% of all advertisement clicks are fraudulent • Packet sniffing What's a botmaster? • Person(s) controlling the botnet o Business person Often paid by customers Willing to rent out botnet o Glory Hound Brags about size of botnet Willing to talk to researchers o Script kiddies Inexperienced Command Topologies • Star o Bots tied to centralized C&C server. • Multi-Server o Same as star but with multiple C&C servers • Hierarchical o Parent bot control child bots • Random o Full P2P support Topology Tradeoffs Control vs. Survivability • More Control o Easier to get botnet to do your bidding o Easier to shut down • Survivability o Harder to shut down o Less control Communication Methods • HTTP o Easy for attacker to blend in • IRC o Harder to hide since IRC is much less used than HTTP • Custom o Makes use of new application protocols Propagation Methods • Scanning o 0-day attacks o Worm-like behavior • Infected e-mail attachments • Drive-by-downloads • Trojan horses Infection Procedure History and Notable Botnets • 1999 - Sub7 • 2000 - GTbot a bot based on mIRC • 2002 - SDbot small c++ binary with widely available source code • 2002 - Agobot staged attacked with modular payload • 2003 - Sinit first peer-to-peer botnet • 2004 - Bagle and Bobax first spamming botnets • 2007 - Storm botnet • 2009 - Waledac botnet • 2009 - Zeus botnet Defense Three main issues: 1. How to find them 2. Decide how to fight them (defense vs offense) 3. How to negate the threat Detection: Analyze Network Traffic • Temporal o Same repeated traffic pattern from node • Spatial o Nodes in same subnet likely infected Detection: Packet Analysis • Using statistical analysis on network traffic flows • Classify packets based on payload signature and destination port o Looking for clusters of similar data packets o n-gram byte distribution • IRC botnet traffic it is not very diverse compared to traffic generated by humans Strategy Active: attack the source • Shut down C&C server • Re-route DNS • Pushback Passive: defend at the target • Filters • Human attestation • Collective defense Defense - Change DNS routing Defender figures out domain that attacker is using and takes control Pros: • Central point of attack • Severs botmaster's ability to communicate with the botnet Cons: • Not all bot nets have C&C server • C&C domain changes often o > 97% turn over per week Defense -Black Lists Defender creates list of attackers. Used primarily as spam fighting technique Pros: • Allows for broad knowledge sharing • Easy to maintain/understand Cons: • List has to be continually updated • Innocent service providers get blocked Defense -Human Attestation Defender requests that client prove his humanity. • Requires the client to have a trusted attester o Accomplished through the use of a Trusted Platform Module • Several methods for an attester to determine that the actions were initiated by a human o Through the use of secure input devices which cryptographically sign their output o CAPTCHA or secure prompt o Analyze keystrokes and mouse movement Defense - Collective defense We must all hang together or assuredly we shall all hang separately. -- Benjamin Franklin • Key contentions o Most end users don't know/care about security o The best way to secure the internet is through a collective effort without relying on end users o Compromised hardware must be quarantined until healthy • Authenticate healthiness before network access o Public Health Model for Internet • Allow everyone but identify suspicious behavior o Japan's Cyber Clean Center o Finnish national Computer Emergency Response Team Thanks